The Committee for Animal Experimentation of Universidad de Caldas approved all the protocols described in this section before starting the trial. All the methods performed in this study were carried out in accordance with The Guide for the Care and Use of Laboratory Animals of the National Research Council and the Institute for Laboratory Animal Research of USA and the Colombian Act of Animal Welfare. Furthermore, this study was carried out in compliance with the ARRIVE guidelines. However, the animals evaluated in this study were Holstein cows from commercial herds. The cows never were slaughtered in the present study for research purposes.
Study design
The study design for this treatment comparison was a non-inferiority randomized clinical trial following a previously described methodology32. The null hypothesis was that the treatment with the experimental treatment (P-PRP) was inferior to the reference product (cefquinome sulfate, CS) administered to the reference group. The alternative hypothesis was that the experimental treatment is not inferior by more than the predefined margin (− Δ):
$${text{H}}_{0} : , left[ {R_{{{text{cure}}}} left( {{text{CS}}} right) , – R_{{{text{cure}}}} left( {{text{P}} – {text{PRP}}} right)} right] , le , – Delta ,$$
(1)
$${text{H}}_{{text{a}}} : , left[ {R_{{{text{cure}}}} left( {{text{CS}}} right) , – R_{{{text{cure}}}} left( {{text{P}} – {text{PRP}}} right)} right] , > , – Delta ,$$
(2)
where Rcure is the cure risk and Δ is the non-inferiority margin. Results were interpreted according to the principles for a non-inferiority trial32,33. Thus, rejecting H0 results in acceptance of Ha, indicating that P-PRP is non-inferior to CS.
Sample size
The sample size calculation was based on previous results using platelet lysates for the treatment of CM, assuming that PRP cure risk was approximately 35%31, and a significance level of 5% and power of 80% were selected. Selection of Δ is often based on the results of negative controlled trials, with Δ being no more than half the effect expected from a superiority study34. However, to our knowledge, there are no negative controlled studies conducted to treat SCM in cows using P-PRP. Therefore, the Δ was chosen as half of the effect of the results reported by Lange-Consiglio et al. (Δ = 17.5%)31. If there is a true difference in favor of the experimental treatment of 5%, then 108 (i.e., 54 cows per group) cows with SCM are required to be 80% certain that the upper limit of a one-sided 95% confidence interval (CI) (or equivalently, a 90% two-sided CI) will exclude a difference in favor of the CS group of more than 17.5%.
Dairy herds and animals
A total of 103 cows from nine commercial herds in Caldas and Risaralda provinces of Colombia were selected for the study. Cows were managed under rotational grazing systems and received supplementation with concentrates according to milk yield. Predominant pastures were a mixed of Kikuyu grass (Pennisetum clandestinum), Orchardgrass (Dactylis glomerata L.), and Yorkshire fog grass (Holcus lanatus). Concentrates used were commercial mixes of cereals, containing between 14 and 16% of crude protein, and approximately 2.9 Mcal ME/kg of dry matter. Concentrates were fed starting at three weeks before calving (2 kg/cow/day), and at a rate of 1 kg per 4 kg of milk yield after calving. Mineral supplements and water were available ad libitum.
Inclusion criteria
Cows for the trial were eligible for inclusion in the study when they were in parity 1 to 5, did not have blind mammary quarters, had experienced no cases of CM in the current lactation, and had received no treatment with antibiotics for any other disease in the last 30 days.
Bacteriological analysis and intramammary infection declaration
Two individual composite milk samples from each cow were aseptically collected before milking at the first visit to the herd in order to select the cows to be treated for SCM. Milk samples were then refrigerated and submitted to our laboratory for analysis. One sample was used for the analysis of SCC using an automated cell counter (Fossomatic, Foss, Hillerød, Denmark). The SCC results were expressed as the natural log (LnSCC, in thousands/mL) to normalize the data distribution.
Based on the SCC results, the second sample was used for bacteriological and polypeptide analyses whenever SCC was ≥ 100,000 cells/mL in primiparous cows and ≥ 200,000 cells/mL in multiparous cows5. Microbiological analyses were performed according to protocols established by the National Mastitis Council35. Briefly, milk samples were streaked onto agar plates containing 5% sheep blood and 0.1% esculin. Agar plates were incubated at 37.5 °C and visually inspected at 24 h and 48 h for bacterial growth. Morphology and Gram staining were used for initial identification. Then a catalase test was used to differentiate staphylococci from streptococci. The coagulase test was performed on catalase-positive cocci to confirm the presence of Staphylococcus aureus. Gram-positive and catalase-negative cocci were differentiated by their reaction to the hydrolysis of esculin under ultraviolet light. The Christie, Atkins, Munch-Petersen (CAMP) test and esculin reaction were used to differentiate Streptococcus agalactiae and Strep. dysgalactiae, while esculin reaction and culture in enterococcosel (Becton–Dickinson, Durham, NC, USA) agar were used for the identification Strep. uberis. Their morphology and Gram staining were used to identify Corynebacterium spp. Growth in MacConckey agar and biochemical tests such as citrate, indole, oxidase and motility tests were used to identify coliforms. Cultures that presented more than two bacterial species were considered contaminated and not informative of intramammary infection (IMI).
Intramammary infection was declared when a composite milk sample presented a SCC higher than the cut points established and the microbiologic culture was positive for any major Gram-positive mammary pathogen (Staph. aureus, Strep. agalactiae, Strep. uberis and Strep. dysgalactiae). All culture-positive cows meeting the SCC selection criteria were treated.
Treatment protocols
Cows with SCM were randomly allocated to one of two treatment groups using random function in Excel (Microsoft Corp., Redmond, WA, USA). A total of 54 cows diagnosed were treated with P-PRP, while 49 cows were treated with CS. The initial treatment was administered 4 days after the first milk sampling. Each quarter of each cow of both P-PRP and CS groups, were treated post-milking and after teat disinfection with the administration of three tubes at 12 h intervals. The P-PRP group was treated with an intramammary infusion of 10 mL of P-PRP activated with calcium gluconate (9.3 mg/mL) (Ropsohn Therapeutics, Bogota DC, Colombia) in a proportion of 1:10; and CS group received the on-label treatment, which consist of 75 mg of CS during three consecutive milkings (Cobactan LC, Merck Animal Health, Madison, NJ, USA).
Blood procurement and P-PRP preparation
The P-PRP used in this study was obtained from four Blanco Orejinegro heifers aged ranged between 12 and 18 months, and mean body weight of 300 ± 20 kg. Animals were sedated with xylazine (0.01 mg/kg, IM) and restrained in right lateral recumbence on a surgical table. Then, the skin of the left lateral jugular vein was aseptically prepared for blood extraction using an iodine povidone scrub and cleared with ethyl alcohol. A small area of the skin over the jugular vein in the central region of the neck was desensitized by the injection of 2 mL of lidocaine (2%). A skin incision of 3 mm was performed with a sterile surgical blade. Then, a 14 G catheter was placed into the jugular vein, and a heparin rubber cap was placed on the catheter to avoid the risk of bleeding.
The blood of each donor was obtained by coupling the needle of a double transfusion bag to the rubber cap. The blood was extracted to the transfusion bag, which was constantly and smoothly shaken to mix the blood and the anticoagulant. A total of three bags of 500 mL were obtained from each heifer during each round of blood extraction. All heifers were closely monitored for anemia (through a weekly complete cell blood count) and any other health issues. All animals were bled every two weeks until treatments in the P-PRP group were complete.
Following extraction, the blood bags were immediately centrifuged in a stationary centrifuge (RotoSilenta 630 RS. Hettich, Tuttlingen, Germany) at 1600g for 8 min. Then, the plasma fraction (P-PRP), including the buffy coat of each blood bag, was transferred to a separated plasma bag. The P-PRP was gently homogenized and packed into 10 mL syringes in a laminar flow cabinet. The free extreme of each P-PRP syringe was protected with a plastic sterile cap. An additional sterile reaction tube containing 1 mL of calcium gluconate was packed into the 10 mL syringe containing the P-PRP in a sterile plastic bag. Calcium gluconate was added to the P-PRP immediately before the intramammary infusion. This substance was used to induce platelet activation and subsequent growth factor release from P-PRP.
Cellular and polypeptide quality control of P-PRP
The platelet and leukocyte counts as well as TGF-β1, PDGF, PF4, and C3 concentrations in whole blood and P-PRG supernatants (P-PRGS) were compared to demonstrate the enrichment of the hemoderivative experimentally evaluated in the present study as follows.
Thirty P-PRP syringes of 10 mL each were randomly chosen for a blood count using automated equipment (Celltacα MEK-6450. Nihon Kohden, Tokyo, Japan). Samples were also analyzed by Enzyme-Linked ImmunoSorbent Assay (ELISA) to determine the concentration of polypeptides released from P-PRGS (i.e., TGF-β1, PDGF-BB, PF4, and C3). P-PRP was incubated with calcium gluconate in a 10:1 dilution for 6 h. The same molecules were measured in plasma (i.e., a negative control) obtained using the same anticoagulant but centrifuged at a higher speed (5000g). We used P-PRP samples incubated with a 0.5% solution of a non-ionic detergent (platelet lysates) (Triton X100, PanReacAppliChem, Barcelona, Spain) in a 10:1 dilution for 15 min as a positive control of polypeptide release, because Triton X100 induces membrane cell damage and complete release of GFs and cytokines from platelets and leukocytes.
Post-admission sampling
Once cows were treated, composite milk samples from both groups were collected at days 21 and 22, and each sample was analyzed for microbiological diagnosis and SCC. Procedures for SCC and microbiology were previously described in “Bacteriological analysis and intramammary infection declaration” section. The treatment was repeated in cows that remained positive for major Gram-positive bacteria, and they received the same product used for the first treatment 4 days after collecting the second milk sample. A final milk sampling was conducted on all refractory cases 21 and 22 days after the repeated treatment (i.e., days 47 and 48). Individual cow SCC, and pre- and post-treatment concentration of cytokines (IL-1, IL-2, IL-4, IL-6, IL-8, IFN-γ, and TNF-α) were generated for each milk sample. No more than two treatment attempts per cow were used in the present study.
Concentration of cytokines in hemoderivatives and milk
Both hemoderivatives and milk samples were thawed at room temperature and used for ELISA analysis without further centrifugation. Mediators were analyzed in hemoderivatives and whole milk using commercial ELISA development kits from R&D Systems (Minneapolis, MN, USA) with the exception of IL-1, IL-4, IL-8 and C3. PDGF-BB (Human PDGF-BB DuoSet, DY220), TGF-β1 (Human TGF-β1 DuoSet, DY240E), and PF4 (Human CXCL4/PF4 DuoSet, DY795) were determined using human antibodies. Notably, TGF-β1 shares a high homology between these proteins in humans and cattle36. Furthermore, human PDGF-BB antibody has been also used to measure the same bovine protein in other studies37,38, while PF4 presents a high homology between humans and ruminants39. Interleukin 1 (Bovine IL-1β ELISA Reagent Kit, Thermo Fisher Scientific Inc., Waltham, MA, USA), IL-2 (Bovine IL-2 DuoSet ELISA. R&D Systems, Minneapolis, MN, USA), IL-4 (IL-4 ELISA Development Kit. Mabtech AB, Nacka Strand, Sweden), IL-6 (Bovine IL-6 DuoSet ELISA. R&D Systems, Minneapolis, MN, USA), IL-8 (IL-8 ELISA Development Kit. Mabtech AB, Nacka Strand, Sweden), IFN-γ (Bovine IFN-gamma DuoSet. R&D Systems, Minneapolis, MN, USA), TNF-α (Bovine TNF-alpha DuoSet ELISA, R&D Systems, Minneapolis, MN, USA), and C3 (Bovine Complement Component 3 ELISA Kit. MyBioSourceInc., San Diego, CA, USA) were assayed with bovine-specific antibodies. The standards provided for each ELISA kit were used in preparing each standard curve according to the manufacturers’ instructions. Readings were performed at 450 nm. In general, the inter- and intra-assay coefficients of variation for the various ELISA kits were between 2 and 5%.
Primary and secondary outcomes
Cure was established in cows that were infected at the beginning of the study and where the organism present was not isolated in the subsequent two post-treatment sample.
Cure was defined at the cow level, and cure risk was statistically assessed for the initial treatment (samples at 21 and 22 days) and treatment of refractory cases samples (samples at 47 and 48 days) independently.
A reduction in SCC was defined as a decrease in SCC for the two post-treatment samples compared to the first value at the start of the study. Changes in milk cytokine concentrations were also used as a criteria for cure and the evaluation of mammary gland inflammation.
Statistical analysis
Platelet and leukocyte counts in whole blood and P-PRP were compared using a Mann–Whitney U Test. The concentrations of TGF-β1, PDGF-BB, PF4, and C3 in hemoderivatives (plasma, platelet lysates, and P-PRGS) were analyzed by one-way ANOVA followed, if necessary, by a Tukey test.
Statistical analysis of the various outcomes was performed using linear and logistic regression models according to the response variable40. The main predictor was the treatment group (P-PRP and CS). The analyses considered covariates such as herd, cow parity, and the natural log of SCC (LnSCC) before treatment. The logistic regression model that was used for bacteriological cure was:
$${text{Logit }}left( {p,{text{of}},{text{ cure}} = {1}} right) , = {text{ intercept }} + {text{ treatment }} + {text{ herd }} + {text{ parity }} + {text{ LnSCC,}}$$
(3)
where cure is the bacteriological cure for major Gram-positive pathogens, treatment is a binary variable indicating either P-PRP or CS, herd is a set of indicator variables for dairy herd, parity is parity of the enrolled cow, and LnSCC is the LnSCC value before treatment. Logistic regression models were also used to determine the probability of cure of Staph. aureus and Streptococcus spp. as outcome variables individually. Streptococcus spp. was a group comprised of the major pathogens of streptococci genera isolated in the study i.e., Strep. agalactiae, Strep uberis, and Strep dysgalactiae.
The non-inferiority trial was obtained by a two-proportion test and based on the upper limit of the 95% CI of the risk of cure, the noninferiority was claimed if the difference in the test was smaller than the Δ41.
The statistical analyses of LnSCC and milk concentration of IL-1, IL-2, IL-4, IL-6, IL-8, IFN-γ, and TNF-α results were performed using linear regression. The main predictor was treatment, and the fixed effects of covariates such as herd, parity, and LnSCC before treatment were also considered. The linear regression model was:
$$y = {text{ intercept }} + {text{ treatment }} + {text{ herd }} + {text{ parity }} + {text{ LnSCC }} + {text{ error,}}$$
(4)
where y corresponds to LnSCC after treatment on day 21 and day 47 for cows treated one or two times, respectively, and the milk concentration of IL-1, IL-2, IL-4, IL-6, IL-8, IFN-γ, and TNF-α, which were measured only at days 0, 21, and 47, where day 0 was the reference value to compare against day 21 and 47 results. The other predictors were previously described.
The analyses were conducted in Stata 14.1 using the commands ‘ranksum’, ‘anova’ ‘prtest’, ‘logit’ and ‘reg’ (Stata Corp. College Station, TX, USA). A P-value of < 0.05 was accepted as statistically significant for all tests.
